Atmos.Chem.Phys.Discuss.,14,28571–28608,2014 D www.atmos-chem-phys-discuss.net/14/28571/2014/ isc doi:10.5194/acpd-14-28571-2014 us s ©Author(s)2014.CCAttribution3.0License. io n P a Thisdiscussionpaperis/hasbeenunderreviewforthejournalAtmosphericChemistry pe r andPhysics(ACP).PleaserefertothecorrespondingfinalpaperinACPifavailable. | Complex chemical composition of D is c u colored surface films formed from ss io n reactions of propanal in sulfuric acid at P a p e upper troposphere/lower stratosphere r | aerosol acidities D is c u s A.L.VanWyngarden1,S.Pérez-Montaño1,J.V.H.Bui1,E.S.W.Li1, sio T.E.Nelson1,K.T.Ha1,L.Leong1,andL.T.Iraci2 n P a 1DepartmentofChemistry,SanJoséStateUniversity,SanJosé,CA95192,USA pe r 2AtmosphericScienceBranch,NASAAmesResearchCenter,MoffettField,CA94035,USA | Received:27October2014–Accepted:28October2014–Published:19November2014 D is Correspondenceto:A.L.VanWyngarden([email protected]) c u s PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. sio n P a p e r 28571 | Abstract D is c u Particles in the upper troposphere and lower stratosphere (UT/LS) consist mostly s s of concentrated sulfuric acid (40–80wt%) in water. However, airborne measure- io n ments have shown that these particles also contain a significant fraction of organic P a 5 compounds of unknown chemical composition. Acid-catalyzed reactions of carbonyl pe r species are believed to be responsible for significant transfer of gas phase organic speciesintotroposphericaerosolsandarepotentiallymoreimportantatthehighacidi- | tiescharacteristicofUT/LSparticles.Inthisstudy,experimentscombiningsulfuricacid D is (H SO ) with propanal and with mixtures of propanal with glyoxal and/or methylgly- c 2 4 u 10 oxal at acidities typical of UT/LS aerosols produced highly colored surface films (and ssio solutions) that may have implications for aerosol properties. In order to identify the n P chemicalprocessesresponsiblefortheformationofthesurfacefilms,AttenuatedTotal a Reflectance–FourierTransformInfraredand1HNuclearMagneticResonancespectro- pe r scopieswereusedtoanalyzethechemicalcompositionofthefilms.Filmsformedfrom | propanalwereacomplexmixtureofaldolcondensationproducts,acetalsandpropanal 15 D itself. The major aldol condensation products were the dimer (2-methyl-2-pentenal) is and 1,3,5-trimethylbenzene, which was formed by cyclization of the linear aldol con- cu s densation trimer. Additionally, the strong visible absorption of the films indicates that s io higherorderaldolcondensationproductsmustalsobepresentasminorspecies.The n P majoracetalspecieswere2,4,6-triethyl-1,3,5-trioxaneandlonger-chainlinearpolyac- a 20 p e etalswhicharelikelytoseparatefromtheaqueousphase.Filmsformedonmixturesof r propanalwithglyoxaland/ormethylglyoxalalsoshowedevidenceforproductsofcross- | reactions.Sincecross-reactionswouldbemorelikelythanself-reactionsunderatmo- D spheric conditions, similar reactions of aldehydes like propanal with common aerosol is c organicspecieslikeglyoxalandmethylglyoxalhavethepotentialtoproducesignificant u 25 s s organic aerosol mass and therefore could potentially impact chemical, optical and/or io n cloud-formingpropertiesofaerosols,especiallyiftheproductspartitiontotheaerosol P a surface. p e r 28572 | 1 Introduction D is c u Aerosolsintheuppertroposphereandlowerstratosphere(UT/LS)arecomposedpri- s s marilyofsulfuricacid(40–80wt%)(Cleggetal.,1998;Finlayson-PittsandPitts,2000; io n Tabazadeh et al., 1997) and water but they also contain significant fractions of or- P a 5 ganiccompounds(Froydetal.,2009;Murphyetal.,1998,2007,2014).Inthecaseof pe r UT aerosols, the amount of organic material can even exceed the amount of sulfate present(Murphyetal.,1998).Thepotentialimpactsofthisorganicmaterialonchemi- | cal,opticalandcloud-formingpropertiesofUT/LSaerosolsarehighlyuncertainsince D is relativelylittleisknownaboutthechemicalcompositionoftheorganicfractionbecause c u 10 availablesamplingtechniquesandfrequenciesarelimitedbythehighaltitudeairborne ssio missionsrequired. n P IncontrasttoUT/LSaerosols,troposphericaerosolsarebettersampledsoitiswell a p establishedthattheycontainmajorfractionsoforganics(upto90%)(e.g.,Calvoetal., e r 2013; Hallquist et al., 2009; Jacobson et al., 2000; Jimenez et al., 2009; Kanakidou | etal.,2005;Murphyetal.,2006;Zhangetal.,2007),andtherehavebeenmanystudies 15 D aimedatchemicalcharacterizationoftroposphericorganicaerosol(OA)particlesand is at determining the physical/chemical pathways for the formation of OA. In particular, cu s reactions of carbonyl-containing organic species including aldol condensation, hemi- s io acetal/acetal formation, organosulfate formation and various polymerization reactions n P have all been identified as potential sources of low-volatility organic products in tro- a 20 p e posphericorganicaerosols(BarsantiandPankow,2004;ErvensandVolkamer,2010; r Gao et al., 2004; Garland et al., 2006; Holmes and Petrucci, 2007; Jang et al., 2002, | 2004; Kalberer et al., 2004; Liggio and Li, 2006, 2008; Liggio et al., 2007; Lim et al., D 2010; Michelsen et al., 2004; Nozière and Esteve, 2007; Nozière and Riemer, 2003; is c Sareen et al., 2010; Shapiro et al., 2009; Surratt et al., 2007, 2006; Tan et al., 2010; u 25 s s Tolockaetal.,2004;Zhaoetal.,2005;ZiemannandAtkinson,2012).Sincethesere- io n actionsarealleitheracid-catalyzedorrequiresulfate,theyarelikelytobeevenmore P a favorableatthehighsulfuricacidconcentrationstypicalofUT/LSaerosols. p e r 28573 | Preliminary experiments for the current work, in which various carbonyl species D is (propanal, glyoxal and/or methylglyoxal) were combined with highly concentrated sul- c u furic acid to simulate UT/LS aerosol acidities, produced highly colored solutions; and ss io solutions containing propanal produced reaction products that partitioned to the liq- n P 5 uidsurfaceasmacroscopicsemi-solidsurfacefilmsthatwerealsohighlycolored.The a p possibility that similar organic products could partition to thin layers or films on the e r surface of UT/LS aerosols is of particular interest because organic compounds that coat aerosol particles would have the most dramatic effects on aerosol chemical, op- | D tical and/or cloud-forming properties (see Donaldson and Vaida (2006) and McNeill is 10 etal.(2013)forreviewsofaerosolsurfacecoatingsandtheirimpactsonaerosolprop- cus erties).Forexample,organiccoatingsonaqueousdropletsandsulfuricacidaerosols s io have been observed to impede water uptake and/or evaporation in laboratory exper- n P iments (e.g., Davies et al., 2013; Otani and Wang, 1984; Rubel and Gentry, 1984; a p Seaveretal.,1992;Xiongetal.,1998),soorganiccoatingsonUT/LSaerosolsand/or er droplets could potentially inhibit water condensation and therefore cloud formation 15 | and/orgrowth.Organiccoatingsmayalsoimpactheterogeneousreactionsataerosol D surfaces;forexample,reactiveuptakeofN2O5hasbeenshowntobeimpededbyvar- isc ious organic coatings which could reduce the rate of hydrolysis of N O to HNO on u 2 5 3 s sulfuricacidaerosols,affectingNOxandOHbudgets(Anttilaetal.,2006;Badgeretal., sion 20 2006;CosmanandBertram,2008;Cosmanetal.,2008;Escorciaetal.,2010;Evans P a andJacob,2005;Folkersetal.,2003;Gastonetal.,2014;Knopfetal.,2007;McNeill p e et al., 2006; Park et al., 2007; Riemer et al., 2009; Thornton and Abbatt, 2005). Sim- r ilarly, organic coatings on sulfate aerosols would alter optical properties, especially if | theorganicsarehighlyabsorbingintheUV-visible.Inordertoassesswhetherspecies D thatformsurfacefilmsonpropanal/H SO mixturesinthelaboratorycouldbeimpor- is 25 2 4 c u tantinUT/LSaerosols,thereactionsresponsibleforfilmformationmustbeidentified, s s whichis,therefore,thefocusofthepresentwork. io n Recent work with various other aldehydes (Li et al., 2011; Sareen et al., 2010; P a Schwier et al., 2010) demonstrated that products of reactions of formaldehyde, ac- p e r 28574 | etaldehyde,glyoxal,methylglyoxalandtheirmixturesaresurfaceactiveeveninwater D is andammoniumsulfate/watersolutionscharacteristicoflessacidiclowertropospheric c u aerosols. Their chemical characterization of the reaction products identified hemiac- ss io etal oligomers and aldol condensation products, but the surface active species were n P 5 notspecificallyidentified. a p In order to identify the chemical species present in films formed by propanal and e r sulfuric acid, we consider the products of the following potential reactions (identified | byletterinFig.1):(A)aldolcondensation,(B)trimethylbenzeneformationviacycliza- D tionofthelineartrimerproducedbyaldolcondensation,(C)hemiacetal,acetal,and/or is 10 polyacetal formation, (D) trioxane formation via cyclotrimerization and (E) organosul- cus fate formation. Each of these processes result in higher molecular weight products, s io whichcouldresultinpartitioningtothesolidphaseasasurfacefilm. n P AldolcondensationproductsareexpectedsinceitcanbeseenfromFig.1thatthey a p aretheonlypotentialproductscontainingsufficientconjugationtoabsorbvisiblelight, er buttheyarenotnecessarilythemajorcomponentofthefilmssinceonlytinyamountsof 15 | suchchromophoresarenecessaryforcolor(McLaren,1983).Productsofpropanalal- D dolcondensationreactionshavebeenobservedinaqueousmediacontainingvarious is c catalysts including anion exchange resin (Pyo et al., 2011), ammonium and carbon- u s s ate salts (Nozière et al., 2010), mixed metal oxides (Tichit et al., 2002), and zeolites io n 20 (Hoangetal.,2010).Inthecaseofzeolitecatalysts,1,3,5-trimethylbenzenewasalso P a observedandproposedtoformfromthelineartrimerproducedbyaldolcondensation p e reactions (Fig. 1b). Aldol condensation reactions of propanal have also been studied r inconcentratedsulfuricacid(60–96wt%)solutionsbyNoziereandEsteve(2007)and | Casaleetal.(2007).NoziereandEstevereportedtheUV-visspectraofaldolconden- D sationproductsof6carbonylcompoundsincludingpropanalandconcludedthattheir is 25 c u absorptionindexcouldbecomesignificantovertheapproximatelytwoyearresidence s s time of stratospheric aerosols. Casale et al. (2007) measured bulk reaction rates for io n a series of aliphatic aldehydes (C – C ), showing that butanal and propanal had the P 2 8 a highestreactionrates,butconcludingthattherateswerenotfastenoughtoberespon- p e r 28575 | sible for transfer of significant organic mass into tropospheric aerosols. Both studies D is focused on aldol condensation reactions due to their potential to form light absorbing c u compoundsandthereforeusedUV-visiblespectroscopyforproductdetection,whichis ss io notsensitivetoproductsoftheotherpotentialreactions(C-E)consideredhere. n P 5 Propanalmayalsoundergoacid-catalyzedreactionswithitshydratedform(diol)to a p formhemiacetals,acetalsandorlinearpolyacetalsasshowninFig.1c.Inadditionto e r these linear species, propanal may also undergo acid-catalyzed cyclotrimerization to | formacyclicpolyacetal(atrioxane)(Fig.1d).Thesereactionshavenotbeenreported D specificallyforpropanalinsulfuricacid,butGarlandetal.(2006)haveshownthatsulfu- is 10 ricacidaerosolsexposedtohexanalvaporcontainedhemiacetalswhileLietal.(2008) cus identified a trioxane in bulk reactions of octanal with sulfuric acid (but not in sulfuric s io acid aerosols exposed to octanal vapor). (In both studies, aldol condensation prod- n P uctswerealsoobserved.)Furthermore,propanalhasbeenshowntoformatrioxanein a p aqueoussolution(Corrochanoetal.,2010)andtoformamixtureofaldolcondensation er products,hemiacetalsandacetalsinthepresenceofananion-exchangeresincatalyst 15 | (Pyoetal.,2011). D Lastly, alcohols may react with sulfuric acid to form sulfate esters (Deno and New- is c man, 1950; Iraci et al., 2002; Michelsen et al., 2006; Minerath et al., 2008; Van Loon u s s and Allen, 2004, 2008; Vinnik et al., 1986), so alcohol species including the diol io n 20 formedbyhydrationofpropanaland/or(hemi)acetals(Surrattetal.,2008)formedfrom P a propanal(Fig.1c)couldreactdirectlywithsulfuricacidtoformorganosulfatessimilar p e tothoseformedbyreactionofglyoxalonsulfuricacidaerosols(Liggioetal.,2005).An r exampleisshownforreactionofthepropanalhydrateinFig.1e. | InthepresentstudywefirstemployacombinationofAttenuatedTotalReflectance– D FourierTransformInfrared(ATR-FTIR)and1HNuclearMagneticResonance(1HNMR) is 25 c u spectroscopiestoidentifythemajorspeciesinthefilmsformedbypropanalonsulfuric s s acidsolutions.Inordertoapproachmoreatmospherically-realisticmixturesoforganics io n andtoaddressthepossibilityofcross-reactionsbetweendifferentcarbonylspecies,we P a alsoexaminedfilmsformedonmixturesofpropanalwithglyoxaland/ormethylglyoxal. p e r 28576 | Finally, we also used UV-visible spectroscopy of the liquid solutions to gain chemical D is insightintotheidentityofthechromophoresandtoillustratetheirpotentialimportance c u forUT/LSaerosolopticalproperties. ss io n P a 2 Experimentalmethods p e r 5 Aftersurfacefilmswerefirstdetectedonsolutionsofpropanal,glyoxal,and/ormethyl- | glyoxal in sulfuric acid during a preliminary study (see Fig. S1 in the Supplement for D photos of typical surface films), controlled survey studies were performed to examine is c u the conditions required for formation of surface films. In these experiments, samples s s of propanal, glyoxal, and/or methylglyoxal in all possible combinations of 1, 2 or all 3 io n 10 species(0.030Mineachorganicpresent)werepreparedinstocksolutionsof19,37, Pa 48and76wt%sulfuricacid(H SO ).AlthoughUT/LSaerosolconcentrationsofthese p 2 4 e organiccompoundsareunknown,0.03MislikelymuchlargerthanUT/LSconcentra- r tionsofanyonecarbonylspecies,butismorereasonableifconsideredasrepresenta- | tiveofthetotalaldehydeorcarbonylconcentration.Sulfuricacidstocksolutionswere D 15 prepared by dilution of concentrated sulfuric acid (96–98wt%, Sigma-Aldrich, ACS iscu grade)withMilli-Qwater,andconcentrationswereconfirmedbytitrationwithstandard- s s ized sodium hydroxide (0.5 N, Sigma-Aldrich). The following Sigma-Aldrich organics ion were used: 97wt% reagent grade propanal, 40wt% glyoxal and 40wt% methylgly- P a p oxal in water. 4.0mL aliquots of each mixture were transferred to multiple 8mL glass e r vialsandstoredundereachofthefollowingtemperatureandlightingconditions:room 20 temperature (21–24◦C)/constant fluorescent light, room temperature/dark, 0◦C/dark, | −19◦C/dark.Sampleswerevisuallymonitoreddailyforcolorchangesandformationof D is surfacefilmsinordertosurveywhichmixturesformedfilmsandtoassesstheimpact c u s ofacidity,organicmixture,temperatureandfluorescentlightonfilmformationrates. s Chemical analysis of the films required production of films in sufficient quantity to ion 25 P allow physical removal of a portion without disturbing the underlying sulfuric acid so- a p lutionsandtherebyavoidingspectroscopicinterferencesfromwaterandsulfuricacid. e r 28577 | Therefore,samplesusedforchemicalanalysiswerepreparedasabove,exceptatthe D is higher concentration of 0.30M in each organic and were stored in volumetric flasks c u (roomtemperature/fluorescentlight)whichcausedthefilmtoconcentrateonthesmall ss io liquidsurfaceareaintheneckoftheflask.Filmsampleswereremovedandtransferred n P 5 with a glass rod to the surface of an ATR crystal for analysis by FTIR spectroscopy. a p ATR-FTIR spectra of the films and standards were collected on a Nicolet 6700 spec- e r trophotometer from 4000–700cm−1 at 1cm−1 resolution using an MCT detector, and a 10-bounce AMTIR ATR crystal with 45◦ mirrors from PIKE. ATR-FTIR was chosen | D forchemicalanalysissincethesemi-solidfilmscouldbedirectlyanalyzedonacrystal is 10 compatiblewithconcentratedsulfuricacidandwithoutanyneedtoalterthechemical cus environment by dissolving the sample in a solvent. In order to provide more chemi- s io cal specificity, films were also analyzed by 1H NMR spectroscopy using a Varian IN- n P OVA400MHzspectrometer.NMRsampleswerepreparedbydissolvingfilmsamples a p in deuterated chloroform (CDCl3) in quartz NMR tubes (5mm outer diameter). ATR- er FTIRand/orNMRspectrawerealsorecordedforthefollowingcommerciallyavailable 15 | standards: 2-methyl-2-pentenal (97wt% Sigma-Aldrich), 1,3,5-trimethylbenzene and D 2,4-diethyl-6-methyl-1,3,5-trioxane(TCI). is c Finally,theUV-visibleabsorptionspectraofsolutions(0.030Mineachorganic)were u s s obtainedusingaVarianCary50BioUV/visiblespectrometerwithadiodearraydetec- io 20 torandquartzcuvettesofvariouspathlengthsfrom0.01–10mmfordifferentregionsof nP a the spectrum. Prior to analysis, solutions were filtered through 2.5µm Teflon filters to p e removeanysuspendedsolidparticulates. r | D 3 Results is c u s 3.1 Formationoforganicsurfacefilms s io n 25 Carbonyl-containingorganics(propanal,glyoxaland/ormethylglyoxal)mixedwithsul- Pa p furicacid(19–76wt%)tosimulateUT/LSaerosolaciditiesproducedcoloredsolutions, e r 28578 | precipitates and surface films. At the highest acidities, all individual organics and or- D is ganicmixturesexamined(0.030Mineachorganic)producedvisiblycoloredsolutions c u that darkened with time. Mixtures containing propanal produced the most deeply col- ss io oredsolutions,progressingfromyellowtoorangetoredtobrownovertimescalesrang- n P 5 ingfromminutestomonths.Thiscolordarkeningprogressedfasterathigheracidities, a p consistent with an acid-catalyzed reaction. Many propanal-containing mixtures also e r eventuallyproducedcoloredprecipitatesand/orsurfacefilms.(Mixturescontainingonly | glyoxaland/ormethylglyoxaldidnotproducesurfacefilms.)Thesesolidsorsemi-solids D were observed either as particles suspended in the liquid (usually collecting near the is 10 surface) and/or as semi-rigid macroscopic films on the surface. In principle, the films cus couldpotentiallybeformedeitherbyheterogeneousreactionsattheair/liquidinterface s io orbyliquid-phasereactionsresultinginproductsthatpartitiontothesurface.Thelatter n P process,however,issupportedbytheobservationthatwhensolutionswerestoredin a p volumetricflaskssolid,darkcoloredmaterialsometimescollectedontheupperslanted er wallsinthebodyoftheflaskbeforemigratingtothesurface;presumablythematerial 15 | rose due to its low density relative to the solution, but was temporarily impeded from D reaching the surface by the flask walls. Furthermore, the quantity of film material ob- is c servedcannotbeeasilyexplainedbyheterogeneoussurfacereactionsalone. u s s Therewasvariabilityinfilmformationratesforreplicatesofthesurveyexperiments, io n 20 however,thefollowinggeneraltrendsemerged.First,theprecisedependenceoffilm- P a formationrateonaciditywascomplex,but,ingeneral,thefilmsformedfasterathigher p e acidity, consistent with acid-catalyzed processes. In fact, the most acidic (76wt% r H2SO4) propanal/glyoxal mixture produced a surface film immediately upon combin- | ingthereactants,althoughotherorganicmixturesformedfilmsmoreslowlyat76wt% D than at 48wt% H SO . Second, film-formation rates also varied as a function of or- is 25 2 4 c u ganic mixture. In general, mixtures containing glyoxal formed films more rapidly than s s thosewithout,whilemixturescontainingmethylglyoxalformedfilmsmoreslowly.Third, io n filmsformedbothinthedarkandunderfluorescentlightwithnoconsistenttrendinfor- P a mationrate.Finally,filmsformedmoreslowlyatcoldertemperatures,but,importantly p e r 28579 | for application to the cold UT/LS, were eventually observed (after approximately 100 D days)evenatthelowesttemperature(−19◦C)examined. isc u s s 3.2 Chemicalcompositionofsurfacefilms io n P a The highly-colored nature of the surface films (only formed on solutions containing p e propanal)isstrongevidenceforaldolcondensationproducts,asaldolcondensationis r 5 the only potential reaction (Fig. 1) of propanal in sulfuric acid that can result in prod- | uctswiththeconjugationrequiredtocauseabsorptionofvisiblelight.Infact,multiple D aldolcondensationstepsarerequiredtoproducesufficientconjugation,sincethefirst isc u aldolcondensationproductofpropanal(2-methyl-2-pentenal,seeFig.1a)iscolorless s s 10 with λmax for the π→π∗ transition of ∼266 and ∼233nm in 75wt% H2SO4 (Casale ion etal.,2007)andwater(ourstandard)respectively.Furtherconjugationfromadditional P a p aldol condensation reactions of 2-methyl-2-pentenal with propanal or with itself is re- e r quiredtoshiftabsorptionintothevisible.Althoughproductsfrommultiplealdolconden- sation steps are almost certainly responsible for the film color, these chromophores | are not necessarily the major chemical components of the films, so ATR-FTIR and D 15 1H NMR spectroscopies were used to analyze the chemical composition of the sur- isc u s face films. The combined results of these two techniques provide evidence that the s io films are a mixture of aldol condensation products (mainly 2-methyl-2-pentenal and n P 1,3,5-trimethylbenzene) and acetals (mainly 2,4,6-triethyl-1,3,5-trioxane and longer- a p chainlinearpolyacetals)asdetailedinSects.3.2.1through3.2.3below.Thedetailed e 20 r chemical analysis in these sections is presented for surface films formed on 0.30M | propanal/48wt% H SO as a starting point, since surface films were only formed 2 4 D on solutions containing propanal and since propanal formed films fastest at 48wt% is c H SO . Sections 3.3–3.4 subsequently address the impact of varying the acidity and u 2 4 s 25 organicmixturefromthisbasecase. sio n P a p e r 28580 | 3.2.1 Aldolcondensationproducts D is c u Figure2presentsatypicalATR-FTIRspectrumofasurfacefilmformedona7dayold s s 0.30Mpropanal/48wt%H SO mixture(ingreen)alongwithspectraoffourstandards io 2 4 n forcomparison.Thestrongabsorptionbandinthefilmspectrumat1689cm−1andthe P a 5 bandat1643cm−1 areconsistentwiththecharacteristicC=OandC=Cstretchingvi- per brations,respectively,ofanα,β-unsaturatedaldehyde,whichisproducedbyaldolcon- | densation (Fig. 1a). The spectrum of neat 2-methyl-2-pentenal shown in Fig. 2 (blue) displays these bands at 1687 and 1643cm−1 and is scaled to illustrate the maximum D is amount of the film spectrum that could be explained by its presence (limited by the cu size of the C=C band at 1643cm−1). An additional C=O peak at 1722cm−1 occurs ss 10 io intheunsaturatedaldehydestretchingregionandisassignedtounreactedpropanal. n P In Fig. 2, the spectrum of neat propanal (red) is also scaled to illustrate its potential a p contributiontothespectrumofthefilm. e r The 1H NMR spectrum for this film presented in Fig. 3 indicates that 2-methyl-2- | pentenal is the dominant species since it contains strong peaks (assigned in Fig. 3) 15 D correspondingtoallfivetypesofhydrogensin2-methyl-2-pentenalinthecorrectmulti- is c plicityandwithin0.03ppmofourstandard.Althoughsomeofthepeaksaretoosmallor u s tooclosetointerferingpeakstointegratereliably,therelativepeakintensitiesarealso sio roughly consistent with the standard. Residual propanal is similarly positively identi- n P 20 fied by comparison to the standard as shown by peak assignments in Fig. 3. There ap e are no additional detectable NMR peaks consistent with linear compounds with ad- r ditional units of conjugation due to multiple aldol condensation steps, indicating that | they must be significantly less abundant than 2-methyl-2-pentenal and therefore will D notcontributesubstantiallytotheFTIRspectrum,either.(Forexample,theprotonsla- is c 25 belled A and B in 2,4-dimethyl-2,4-heptadienal (Fig. 1) would be expected to appear us s as singlets with chemical shifts near those for 2,4-hexadienal (Spectral Database for io n OrganicCompounds,2014,SDBS)at9.5and7.1ppm,respectively.) P a p e r 28581 | Although there is no NMR evidence for linear aldol condensation products beyond D is 2-methyl-2-pentenal,NMRpeaksat2.27and6.79ppmconfirmthepresenceof1,3,5- c u trimethylbenzene (mesitylene) (SDBS), which has previously been observed to form ss io in reactions of propanal over acidic zeolite catalysts (Hoang et al., 2010). Hoang n P 5 et al. proposed that 1,3,5-trimethylbenzene was formed by acid-catalyzed cyclization a p andsubsequentdehydrationofthetrimerformedbyaldolcondensation(2,4-dimethyl- e r 2,4-heptadienal)ofpropanalasshowninFig.1b,whichisareasonablemechanismfor | sulfuricacidsolutionsaswell.Furthermore,thetrimerformedbyaldolcondensationof D acetonehasalsobeenshowntocyclizetoform1,3,5-trimethylbenzeneinsulfuricacid is 10 (Duncanetal.,1998;Kaneetal.,1999;Klassenetal.,1999).InFig.2,theATR-FTIR cus spectrumofneat1,3,5-trimethylbenzene(black)isscaledtothefilmspectrumtoindi- s io cateitsmaximumpotentialcontribution.Thecomparisonshowsthatthefilmspectrum n P isconsistentwiththepresenceofsome1,3,5-trimethylbenzeneinthefilmsinceithas a p bands corresponding to the 2 most intense 1,3,5-trimethylbenzene bands at 834 and er 1609cm−1, the latter of which lies in the region for aromatic skeletal vibrations and, 15 | therefore,cannotbeexplainedbyanyotherpotentialproducts. D Although2-methyl-2-pentenaland1,3,5-trimethylbenzeneareshownheretobethe is c major products resulting from aldol reactions, both are colorless, so the more highly u s s conjugated compounds formed by further aldol condensation steps that are presum- io n 20 ably responsible for the film color must be minor constituents. Therefore, there must P a also be additional compounds present in the film to explain the strength of the peaks p thatappearintheFTIRspectrumbetween1500–800and3000–2800cm−1. er | 3.2.2 Ethers:acetals/hemiacetalsandlinear/cyclicpolyacetals D is c Inadditiontoaldolcondensationproducts,theFTIRandNMRspectrabothalsodisplay u s 25 evidenceforethergroups(C-O-C)duetostrongpeaksinthe1200–1000cm−1 region sio oftheFTIRspectrum(Fig.2)andpeaksinthe4.5–5.1ppmregionoftheNMRspec- n P trum.Speciesthatcouldberesponsiblefortheseethersignaturesincludehemiacetals, a p e acetals and/or higher order polyacetal polymers which can form from the reaction of r 28582 | propanal with one or more of its hydrates (diols) (Fig. 1c) or from cyclo-trimerization D is of propanal to form the cyclic acetal, 2,4,6-triethyl-1,3,5-trioxane (Fig. 1d). Of these c u potentialproducts,thecyclotrimerismosteasilyconfirmedsinceitisreadilyidentified ss io by comparison of the FTIR and NMR spectra of the film to reference spectra (SDBS) n P 5 of2,4,6-triethyl-1,3,5-trioxaneasindicatedbythepeaksassignedtothetrioxane(T)in a Figs. 2 and 3. Specifically, the 1H NMR spectrum of the film contains all three of the pe r peaksinthereferencespectrum:atripletat4.78ppm,acomplexmultipletat1.67ppm | andatripletat0.94ppm(althoughthebroadpeakgroupat0.94ppmcanonlybepar- D tiallyduetothetrioxaneduetoitsstrongintensityrelativetotheothertrioxanepeaks). is 10 Sofimthielarfillym, acsorsrheoswponnbdywaitshsiingn2mcmen−t1stionpFeiga.ks2,inatthleearsetfe1r3enpceeaskpseinctrthuemFoTfIaRnsepaetclitqruumid cuss io filmof2,4,6-triethyl-1,3,5-trioxane(includingall6ofthestrongestreferencepeaksbe- n P tween1500–900cm−1).(Thistrioxaneisnotcommerciallyavailable,soinordertoalso a p provideageneralideaofrelativepeakintensitiesexpectedintheATR-FTIRspectrum er ofthetrioxane,thespectrumofatrioxanethatdiffersonlybyreplacingoneethylgroup 15 | withamethylgroup(2,4-diethyl-6-methyl-1,3,5-trioxane)isalsoshowninFig.2.)Fur- D thermore,previousstudiesof2,4,6-triethyl-1,3,5-trioxanereportthatitphaseseparates is c uponformationfrompropanal/catalystsolutions(Satoetal.,1993),consistentwithour u s s surfacefilmformation. io n 20 Upon assignment of the cyclotrimer peaks, only one major peak in the FTIR spec- P a trum of the film remains unexplained by species identified thus far (2,4,6-triethyl- p e 1,3,5-trioxane, 2-methyl-2-pentenal, 1,3,5-trimethylbenzene and propanal). This peak r at945cm−1 is,however,thestrongestpeakinthespectrumofthefilmandtherefore, | must be a major peak in the spectrum of the absorbing species. The hemiacetal and D singleacetalformedbypropanal(Fig.1c)areunlikelytoberesponsibleforthepeakat is 25 c 945cm−1sincetheywouldbeexpectedtoproducetheirstrongestbandsathigherfre- us s quencies. Specifically, the hemiacetal would produce a strong FTIR absorption band io n in the 1150–1085cm−1 region from the asymmetric stretch of its single ether group, P a whiletheacetalcontainstheC-O-C-O-Cmoietywhichwouldproduce5characteristic p e r 28583 | bands between 1200–1020cm−1 (Bergmann and Pinchas, 1952). Furthermore, both Dis the hemiacetal and acetal are also unlikely to be major ether constituents since only cu a weak peak exists in the OH stretching region (3500–3400cm−1) where a stronger ss io peak(withrespecttothepeaksintheetherregion)wouldbeexpectedduetotheOH n P groups. a 5 p Instead,thestrongpeakat945cm−1 mostlikelyresultsfromlonger-chainpolymers er ofpropanal(polyacetalinFig.1c)dueto-C-O-C-O-C-stretchingbandsthatareshifted | tolowerfrequencieswiththeadditionofadditionalethergroups.Spectraofpolymers D ofvarioussmallaldehydes(formaldehyde,acetaldehydeandpropanal)whichcontain is c thesamepolymethoxy(-C-O-) backbonedisplayonlyveryweakOHstretchingbands u 10 n s butmultipleverystrong,broad,overlappingabsorptionbandsbetween925–975cm−1 sio (NovakandWhalley,1962,1959a,b;Vogl,1964a,b).Althoughthepeakat945cm−1 nP does not exactly match any of the three strongest peaks (975, 960 and 925cm−1) ap e in this region in the Novak and Whalley (1959a) spectrum of the polymethoxy poly- r merformedbypressurizationofpropanal,thereissuchbroadabsorptionintheentire | 15 980–920cm−1 regionofthepolymerspectrumthatapeaknear945cm−1 maynotbe D distinguishable.Furthermore,thepolymerpresentinoursurfacefilmislikelytodisplay isc different relative intensities of the peaks in this region due to differences in degree of us s polymerizationand/ordifferencesinrelativequantitiesofrotationalisomers(Novakand io n Whalley,1959a).Additionally,bandsmayalsobeshiftedinfrequencyduetodifferent P 20 a interactionsbetweenpolymerchains(NovakandWhalley,1959a)inthecomplexsur- pe r facefilmmatrix.Finally,theNMRspectrumofthefilmisalsoconsistentwiththepres- ence of propanal polymer since the 4.5–5.1ppm region contains multiple unassigned | peaksconsistentwithethersandsimilartothebroadgroupofunresolvedpeaksfrom D is 4.5–5.0ppmthatcharacterizestheNMRspectrumofthepolymethoxypolymerformed c 25 u byacetaldehyde(Vogl,1964a),whileCH2 andCH3 protonsfromtheethylchainsare ssio likelyresponsibleforpeaksinthe1.0–1.7ppmregionandforaportionofthetripletat n P 0.94ppm,respectively. a p e r 28584 | Afterthisidentificationofpolymersofpropanal,wenotethatallofthemajorbands D is in the infrared spectrum that could not be explained by aldol condensation products c u either correspond to 2,4,6-triethyl-1,3,5-trioxane or could reasonably be assigned to ss io longer-chainlinearpropanalpolymers. n P a 5 3.2.3 Otherpotentialfilmcomponents:organosulfates,etc. pe r Inordertotestforthepresenceoforganosulfates,reactionmixtureswerepreparedwith | hydrochloricacidatthesamepHasthesulfuricacidmixtures.Theformationofsurface D films on these mixtures demonstrates that organosulfates are not necessary for film isc u formation,andthesimilarityoftheATR-FTIRspectraforfilmsformedonsulfuricacid s s 10 andhydrochloricacidsolutions(exampleshowninFig.4for0.30Mpropanal/48wt% ion H SO )demonstratesthatorganosulfatesarenotpresentinsignificantquantities.We P 2 4 a p note,however,thatorganosulfatescouldstillbeproducedinthesulfuricacidsolutions, e r wheretheywouldbeexpectedtoremainduetotheirpolarnature. Although all of the major peaks (and many of the minor peaks) in both the FTIR | andNMRspectraofthefilmcanbeassignedtothechemicalspeciesdiscussedthus D 15 is far, some small unassigned peaks (e.g. NMR peaks at 3.2 and 3.9ppm) indicate the c u s presence of other minor species. These could include products of multiple aldol con- s io densationsteps,aldolsthathavenotlostwaterthroughthecondensationprocess(see n P Fig.1),thehemiacetalandacetalformedbypropanal,otheracetalsthatcouldalsopo- a p tentially be formed by reactions of aldol condensation products with propanal, and/or e 20 r productsfromoxidationoffilmsbylight/air. | 3.3 Effectofacidity D is c u As discussed in Sect. 3.1, acidity has a complex effect on the formation rates of the ss surfacefilmsthatvariesdependingontheorganicmixture.Ingeneral,filmstendedto ion 25 formfasterastheacidityincreasedfrom19to37to48wt%H2SO4,butfilmsformed Pa p more slowly or not at all at the highest acidity (76wt% H SO ) in all mixtures except e 2 4 r 28585 | propanal/glyoxal.The FTIRspectraoffilms formedonmixtures of0.30Mpropanalin D is 48 and 37wt% H SO solutions in Fig. 5 (shown in triplicate) demonstrate that there c 2 4 u are also chemical differences in films formed at different acidities. The spectra are ss scaled to the C=O peak at 1690cm−1 from aldol condensation products (predomi- ion 5 nantly2-methyl-2-pentenal)inordertoillustratedifferencesinrelativepeakintensities. Pa p Althoughthereisconsiderablevariabilityinrelativepeakintensitiesamongthespectra e r ofreplicates(mostlikelyduetofilminhomogeneity),thepeaksinthe1200–900cm−1 region are generally larger relative to the C=O peak at the higher acidity (red), in- | D dicating a larger relative contribution to the film from the 2,4,6-triethyl-1,3,5-trioxane is 10 afonrdmelodngaetrt-hcehahiinghpeorlyaaccideittaylaplosloymhaevrsetshmatalalebrsoprebakinstahtis16re0g8iocnm.−I1n,aindddiictiaotnin,gthsemfialmllesr cuss io concentrationsof1,3,5-trimethylbenzenerelativetoaldolcondensationproducts.Both n P ofthesetrendsareconfirmedbyNMRspectroscopy(datanotshown).Thepresence a p ofmore2,4,6-triethyl-1,3,5-trioxaneandpolymersathigheraciditiesisconsistentwith er faster film formation at higher acidities in the 19–48wt% H SO range since these 15 2 4 | speciesaremostlikelyresponsibleforthephaseseparationintoasurfacefilm.Addi- D tionally, slower film formation at the highest acidity (76wt%) is potentially due to low is c watercontentthatreducestheformationofthediolsrequiredtobeginthepolymeriza- u s s tionprocess(seeFig.1c). io n P 3.4 Cross-reactionswithglyoxalandmethylglyoxal a 20 p e r To examine the potential effect of additional organic species with carbonyl groups on | theformationoffilms,mixturesofpropanalwithglyoxaland/ormethylglyoxalwerealso D examined. Although glyoxal and methylglyoxal did not form films in the absence of is c propanal,themixturesofpropanalandglyoxalformedfilmsfasterthanpropanalalone, u s 25 whilemixtureswithpropanalandmethylglyoxalformedfilmsmoreslowly.Acomparison sio ofFTIRspectraoffilmsformedonvariousorganicmixturesin48wt%H SO allpre- n 2 4 P paredonthesamedayareshowninFig.6.(Becausesignificantvariabilityinrelative a p e peakintensitiesexistsinreplicateFTIRspectraofthefilmsmostlikelyduetoinhomo- r 28586 | geneity in the solid mixtures of multiple chemical species, a complete set of replicate D spectra are provided in the Supplement (Fig. S2) to demonstrate that the differences isc u between organic mixtures discussed here are in fact due to differing chemical path- ss waysandarenotsimplysamplingartifacts.)ThespectraareagainscaledtotheC=O ion 5 peakat1690cm−1 frompropanalaldolcondensationproductsinordertoillustratedif- Pa p ferencesbetweenrelativepeakintensities.Thespectraofthefilmsfrompropanaland e r propanal/glyoxalarenearlyidenticalinthe1800–1600cm−1regionindicatingthatboth | filmsinclude2-methyl-2-pentenalandunreactedpropanalinsimilarratios.Conversely, thespectralpatterninthe1200–900cm−1regionforthepropanal/glyoxalfilmdoesnot Dis 10 correspondtothespectrumof2,4,6-triethyl-1,3,5-trioxaneasitdoesforthepropanal- cus onlyfilm.Sinceglyoxaldidnotformfilmsbyitself,thisinfraredsignatureismostlikely s io duetoproductsofcrossreactionsbetweenpropanalandglyoxal. n P TheFTIRspectrumofthefilmfrompropanalandmethlyglyoxaldeviatesevenfarther a p from that of propanal-only. Not only does it lack the signature of 2,4,6-triethyl-1,3,5- er trioxane, indicating products of cross-reactions, but, additionally, absorbance in the 15 | entire1500–900cm−1 regionismuchstrongerrelativetothealdolcondensationpeak D at1690cm−1,indicatingastrongerrelativecontributionfromacetalspecies.Finally,itis is c intriguingthat(1)thespectrumofthefilmformedonthepropanal/glyoxal/methylglyoxal u s mixture is quite similar to that for the propanal/methylglyoxal mixture, differing only in sio n 20 relativepeakratiosand(2)thattherateoffilmformationwasdecreasedfromtherate P a forthepropanal/glyoxalmixture.Thiscouldindicatethatglyoxalissomehowinhibited p e fromparticipatinginfilm-formingreactionsbythepresenceofmethylglyoxal. r | 3.5 UV-visspectraofsolutions D is c Althoughthefocusofthisworkisoncharacterizationofthesurfacefilms,theUV-visible u s 25 absorption spectra of aged organic/sulfuric acid solutions were also examined in or- sio der to give some insight into the potential formation of highly absorbing species over n P the long residence time of lower stratospheric aerosols (∼2years). Figure 7a (blue) a p e shows that a solution of 0.030M propanal in 48wt% sulfuric acid allowed to age for r 28587 | 274dayshadtwostrongabsorptionpeaksaround200and245nmmostlikelycorre- D is sponding to species also observed in the films: 1,3,5-trimethylbenzene and 2-methyl- c u 2-pentenal,whichabsorbinwaterat∼200and234nm,respectively.Moreimportantly, ss io theabsorbanceextendssignificantlyintothevisible.Therearenootherdistinguishable n P 5 peaks,buttheabsorbanceismostlikelyduetooverlappingpeaksfromvariouslonger a p oligomers formed by additional aldol condensation reactions of 2-methyl-2-pentenal e r with propanal and/or with itself. Each sequential aldol condensation step would add | another unit(s) of conjugation and thereby shift the absorption peak to longer wave- D lengths.Thisinterpretationissupportedbytheobservationthatwhentheaciditywas is 10 increased to 76wt% sulfuric acid (Fig. 1b), the intensity of the peak corresponding cu s to 2-methyl-2-pentenal was reduced (or even absent) and additional peaks became s io distinguishableatlongerwavelengths(270,365,388(shoulder)and458nm).Nozière n P and Esteve (2007) observed a similar spectrum for reaction products of propanal in a p 96wt%sulfuricacid,andalsoascribetheselongwavelengthpeakstooligomersfrom er aldol condensation reactions. Although they suggest that the peak in their spectrum 15 | near270nmmaybepropanalitself,thiscannotbethecaseforoursamplessincethe D molarabsorptivityofpropanalistoosmallat∼9cm−1M−1 (Xuetal.,1993). is c The absorption spectra of mixtures of propanal with glyoxal and/or methylglyoxal u s arealsopresentedinFig.7.“Effective”molarabsorptivitiesarecalculatedbasedonly sio n 20 on the concentration of the propanal reactant (0.030M) so that any changes in ab- P a sorbance (compared to the propanal-only spectrum) must be due to the presence of p e the additional organic species. At both acidities absorbance in most of the spectrum r is increased, with methylglyoxal having a larger effect than glyoxal, suggesting that | theaddedorganicspeciesareundergoingaldolcondensationeitherviareactionswith D propanalor self-reactions.Althoughsome oftheadditional absorptionmay bedueto is 25 c u glyoxal and methylglyoxal themselves, molar absorptivities of these species are too s s small (Horowitz et al., 2001; Malik and Joens, 2000; Plum et al., 1983) to contribute io n significantlyatleastbelow350nm. P a p e r 28588 | 4 Discussionandatmosphericimplications D is c u Themajorspeciespresentinsurfacefilmsformedonbulksolutionsofpropanalinsul- s s furic acid were identified as aldol condensation products (mainly 2-methyl-2-pentenal io n and 1,3,5-trimethylbenzene) and polyacetals (mainly 2,4,6-triethyl-1,3,5-trioxane and P a 5 longer-chainlinearpolyacetals).Oftheseproducts,thepolyacetalspecies(bothcyclic pe r andlinear)aremostlikelytobeprimarilyresponsiblefortheseparationoftheorganic speciesfromthesolutionintoaseparatesolidorganicphaseonthesurfaceoftheliq- | uidduetotheirhighmolecularweightandhigherhydrophobicitycomparedtothetwo D is observedaldolcondensationproducts.Sincethesolidmaterialinthelaboratorysam- c u 10 plesrisestothesurfaceofthesolutiondue,atleastinpart,toitslowdensityrelative ssio tosulfuricacid,ifsimilarinsolubleacetalswereformedfromreactionsofaldehydesin n P liquid UT/LS aerosols, it is unclear whether they would exist as solid inclusions or as a p surfacecoatings(fullorpartial),thelatterofwhichwouldbemorelikelytoalteraerosol e r optical,chemicaland/orcloud-nucleatingproperties. | Neither the solubility nor the reactive uptake coefficient of propanal in sulfuric acid 15 D has been measured, but, based on the low concentration of propanal vapor in the is UT/LS(∼15pptat11kmandpresumablymuchlowerinthestratosphere,Singhetal., cu s 2004)uptakeandreactionofpropanalalonetoformpolyacetalsisnotexpectedtobe s io asignificantsourceoforganicmaterialinUT/LSaerosols.However,polyacetalforma- n P tionfromaldehydesingeneralcouldbeimportantforthreereasons.First,polyacetals a 20 p e may be formed from a variety of organic species since they have been observed to r formfrommanyaliphaticaldehydes(Vogl,2000)andhavespecificallybeenobserved | in sulfuric acid for formaldehyde uptake (Iraci and Tolbert, 1997) and inferred for ac- D etaldehydeuptake(Williamsetal.,2010).Second,therateoffilmformationwasgreatly is c enhancedbythepresenceofglyoxal,suggestingthatcarbonylspeciesalreadypresent u 25 s s inaerosolscouldenhancethereactiveuptakeandpolyacetalformationofsmallalde- io n hydes,consistentwithpreviousexperimentsthatdemonstratedenhancedreactiveup- P a takeofacetaldehydeonsulfuricacidsolutionscontainingformaldehyde(Williamsetal., p e r 28589 | 2010)andenhancedreactiveuptakeofnonanalonmixedorganic/sulfuricacidaerosols D is (Chan and Chan, 2011). Third, although aerosol concentrations of any one aldehyde c u are unlikely to result in significant self-polymerization, cross reactions between alde- ss io hydesand/orbetweenaldehydesandalcoholsmaybesignificantandarespecifically n P 5 shownheretooccurbetweenpropanalandtwocommonaerosolorganicspecies(gly- a p oxalandmethylglyoxal). e r Althoughuptakeanddissolutionofaldehydesontosulfuricacidaerosolsisthemost | likely method of polyacetal formation directly from small volatile mono-aldehydes like D propanal, there may be more favorable methods for polyacetal formation in UT/LS is 10 aerosols.Sincepolyacetalformationrequiresmultiplepolymerizationsteps,thekinet- cus icsarelikelytobegreatlyenhancedathigherconcentrationsoftheorganicreactants. s io One possibility for enhanced concentration of organic reactants is the potential pref- n P erence of the reactants for the aerosol surface. If polyacetals partition to the aerosol a p surfaceastheydoinourbulkexperiments,theirfurtherpolymerizationwitheachother er and with condensing organics would be enhanced; polymerization could be similarly 15 | enhancedifcarbonyland/oralcoholreactantspartitiontotheaerosolsurfaceortoor- D ganic inclusions. This possibility is supported by the recent work of Li et al. (2011), is c Schwieretal.(2010)andSareenetal.(2010)demonstratingsurfacetensiondepres- u s s sionbysurface-activespeciesformedinsolutionsofformaldehyde,acetaldehyde,gly- io n 20 oxal and/or methylglyoxal in pure water and/or aqueous ammonium sulfate. Addition- P a ally,productsofcross-reactionsbetweenmethylglyoxalandformaldehydeoracetalde- p e hyde had a larger effect on surface tension than could be explained by self-reactions r alone. | Anadditionalpossibilityforenhancedconcentrationsoforganicreactantsfavorable D forpolyacetalformationistransportoforganic-richaerosolsfromthelowertroposphere is 25 c u totheUT/LS.Polyacetalformationcouldbeinitiatedonsuchaerosolsuponcondensa- s s tionofH SO and/orcoagulationwithH SO particlesformednearthetropopause.In io 2 4 2 4 n order to evaluate the likelihood of this process, carbonyl species more typical of pho- P a tochemically aged tropospheric aerosols (less volatile and likely more oxidized than p e r 28590 |
Description: